A [) c. JoH RALPH
Transcript of A [) c. JoH RALPH
Studies in Avian Biology o. 29: 1- 6
USE OF MIST NETS AS A TOOL FOR BIRD POPULATION MONITORING
ERICA H. Du A [) c. JoH RALPH
Ahstract. Mist nets are an important tool for population monitoring. here defined as asse1.,1.,ment of '>pecics composition. relative abundance. population siLe. and demography. We re\ ie\\ the strength<., and limitation-. of mist netting for monitoring purposes. based on papers in this volume and other literature. Ad\antages of using mi'>t nets over aural or\ isual count method include ea1.,e of '>tandardized sampling, low observer bias. ability to detect species that arc often missed using other count rm:thods. and opportunity to examine birds in the hand (providing information on conJition, age. sex. and capture history). The primary limitation of mist netting. in common with most other survey methods. is from potential bias in '>ampling. However. there are many approaches to reducing or adJU'>ting for bias, including standardiz..ition of netting method.,. combining mi.,t-net '>ampling \\Ith other sune) types, and us111g mark-recapture techmque<., . Mist netting is an essential tool for '>pecies inventory. pro\idcs useful indice1., of relative abundance. and can be u<,ed lo track temporal trend.., in abundance It is al. o one of the most efficient methods of capture for mark recapture studie ....
Key Word\: mark- recapture. mist net, population monitoring, sampling bias.
Mist netting is an important technique for population monitoring , helping to a<.,sess specie" composition. relative abundance. population <.,itc. and demography (productivity and survi' al). Whereas
because it recommends best practices in mist netting. accompanied by the reasons why recommended procedure" \\ill improve monitoring capability.
mist netting is time intensive and requires special- SPECIE MPOSITION i1ed training. it ha'> certain achantages over visual and aural population monitoring technique'> Mi"t nets can <;ample species that arc poorly detected by other means. counts are not subject to observer bias. netting effort i.., easil] standardi1ed. and each bird counted can also be examined in the hand. apture allows bird'> to be aged. sexed. and marked to allO\" ind1\ idual id ntill ·~Hion in futurl: enc )linter-.. ln aJ dition. cxt1a data can he collected that al'>o contrihute to population -.tudie. , such as breeding ·tatus or ub-..,pecies identification. Data can be collected f'or
other research purpose<; at the same time (e .g .. phys1-ological state. molt, parasite load..,, DNA sampling). Because 1nist netting is one of the most cf11cient means ot capturing many bird species, especial!) those that are in..,ecti\orous. the technique is often u~ed in mark-recapture studie<; .
In thi paper, we discuss the strengths and limitations of mist netting for population monitoring applications, and summarize the literature in which population parameters based on mist-net captures were evaluated by comparing them with data from independent data sources. In addition. 'We reviev ... the main sources of potential bias in population indices based on numbers of birds captureJ, and di cuss some ways to address such bias. Ralph et al. (!hi volume a) hould be regarded as a companion paper to this one.
Mi1.,t netting is often used as a tool to determine \\hat species are pre. ent in a tud) area. The technique is a valuable component or specie~ inventory becau<;e it detects more cryptic. ground-foraging. and non-singing birJ than aural or \isual suncys (Blak anJ Loiselle 200 I, Rappole et al. l 993, \998, W.1\1<1ce ct .ti \l)96, Whitman ct al. 19)7). Further, rc..,ult'> are relati\ ely unaflected by the bird identification ..,kills of observers (Karr l 981 a; although m1s1dent1fication may still occur, Dale thi\· volume). However. netting is often a less efficient means of species inventory than censuses <,uch as point counts. in term<., of species detected per unit effort (Ralph et al. 1995. Gram and Faaborg 1997, Whitman et al. 1 Q97). Moreover, netting is known to under-sample or completely miss some species (<.,uch as aerial foraging swallows, or raptors). regardless of season (Wang and Finch 2002). As a re..,ult. most authors ha\e recommended that mist netting be used as a supplement to visual or aural surveys when a species inventory is being prepared. rather than as a sole '>Ource of data (Faaborg et al. this volume.
Whitman this \'Olume). Kendall et al. (this 11ol11me)
provide information on u ing mark-recapture techniques to estimate the total species present, even though only a proportion has been detected.
2 STUDIE IN VfA BIOLOGY NO. 29
RELATIVE ABUNDANCE AND TRENDS
Mist-netting studies are commonly used to document differences in abundance indices among species, locations, years, or age classes (see nex section), and to detect trends in population indice~ over the long term. No matter what count method~
are used to obtain abundance indices, the proportion of the true population that is counted will likely var_' over time and pace, introducing bias, which w discuss below. Nonetheles. , evaluation studies have shown that abundance ind ice · derived from mist-net
ampling often compare well to independent data on the parameter. of interest.
For example, species rankings based on relative abundance in breeding season mist-net sample
were usually correlated with abundance ranking based on point counts at the same locations (Table I), although individual species' rankings sometime · differ d markedly between count types (DeSante et al. this volume, Kaiser and Berthold this l'Olume ) .
Similar studies in wintering areas gave mixed re
sults, in that agreement of species' rankings between methods was quite good for some data sets (e.g., Wallace et al. 1996 this 1•ol11111e), but very poor in others (Blake and Loiselle 2001). Faaborg et al. (thi\'
volume) found good correspondence for year-round residents but very little for wintering species, anJ Lynch ( 1989) found that level of correc;pondence differed among habitats. In the migration season, birds are perhaps less selective of specific habitat types (Moore t al. 1995). For example, Wang and inch (2002) found good correspondence between mist-net and point-count abundance rankings of species dur
ing migration in all habitats studied. Within species. annual abundance indices have
been shown to fluctuate in parallel \\ ith indices based on other data sources (Table 1). Repeated mi<.,t netting throughout the breeding season gave indices that paralleled abundance data derived from spot mapping. in 3 of 4 species studied by Silkey et al. ( 1999, from a single netting station) and in 9 of 21 species studied by Peach et al. (this 1•ol11me. pooling data from many locations). No comparable studies have been conduct d during the wintering season. For the migration season. Dunn et al. (this mlume
a) showed that annual abundance indices ba ed on daily mi<,t-net samples were strongly correlated with
indices based on a standardized daily census in 73% of 64 species.
Several comparisons have been made between long-term trends in abundance indices based on netting data and trends from independent sources (Table
l ). Pooled data from constant-effort mist netting
at many locations during the breeding season corresponded with regional population trends based on spot mapping in 15 of 21 species (Peach et al. 1998, this volume) . Trends in numbers of migrants captured
were often correlated with Breeding Bird Surve.> trends from regions to the north where the migrants were assumed to have originated (Hagan et al. 1992. Dunn and Hussell 1995, Dunn ct al. 1997, Francis and Hussell 1998. Berthold this l'Olume, Rimmer et al. this volume). Correlations were strongest when statistical
technique. were used that compensated for variation in daily bird numbers caused by weather and date in the . eason, and preci:ion of long-term trends has been shown to improve when netting at a single station is
more frequent (Thomas et al. this volume). However. as noted by Rimmer et al. (this l'Olume), birds from diverse portions of the breeding range are typically sampled at a single location, making direct comparisons between mist-net capture rates and Breeding Bird Survey trends difficult.
DEMOGRAPHIC MONITORfNG
Monitoring of productivity is a special case of abundance monitoring, in \Nhich abundance of adult and young birds is assessed separately. Because capture probabilities differ between age classes (Ballard et al. this \'Olume, Burton and De ante this l'Olume.
ur et al. this \'Olume). the relative proportions or young to adults cannot be regarded as absolute measures or the number or young produced per adult. but rather are indices of productivity (Bart ct al. 1999). Productivit} indices from constant-effort mist netting in the breeding season have been compared to
the numbers of nestlings found during intensive nest monitoring (Table I). In some. but not all species, these estimates fluctuated in parallel bet\\een year'> (Nur and Geupel I 993b, du Fcu and McMccking this 11olume). Discrepancies may have resulted from postfledging dispersal of young (e.g., Anders et al. 1998. Vega Rivera et al. 1998). so that mist-net sample.., represented local producti ity in some species and regional productivity in others. Differences in mistnet based productivity indices among stations within a region (as found by Ralph et al. this l'Olume b)
could therefore result from true differences in local
productivity, or from post-fledging redistribution of birds. Therefore, unles. pilot work has demonstrated that productivity indice. from mist netting accurately
reflect local productivity in the target species, sitespecific indice. of productivity based on mist netting should at least be augmented by intensive nest monitoring (e.g., Gates and Gysel 1978, Roth and Johnson 1993).
MIST NETS AS A MONITO ING TOOL-Dunn and Ralph 3
In contrast, it ha<> been demonstrated that collecting data from multiple netting stations is a good means of tracking regional productivity (Bart et al. 1999; Table 1 ). Cooperative programs that pool
data from constant-effort sampli ng at many mistnet stations in a region include MAPS (Monitoring Avian Productivity and SurvivoL hip; DeSante et al. this l'Olume), the British Trust for Ornithology's
TABLL J. COMP\RISON Of' POPL.LATION DATA COLITC'1ED 8) MIST'il·TTl"IG WITH DATA FROM INDEPE DENT SOURCES
Para11ctcr
Rel;rive abundance of species
Annual abundance indices for individual species
Dail~ abundance indices
Population trends
L >ea productivit)
Season
Breeding
Winter
Migration
Breeding
Migration
Migration
Migration
Breeding
Migration
Migration
Breeding
Breeding
R 'gional productivity Breeding
Source of data
for comparison
Point counts
Point counts
Point counts
Spot mapping
Transrct
Point counts
Radar
Spot mapping
pot mapping
Breeding Bird urvey
est monitoring
e-;t monitoring
e-.t monitoring
~nrrespondence of
para c1cr beiwcen daia \els Source
orrelated at 34 DeSante et al. this l'Olume, Kaiser and Jf 37 locations Berthold !hi\ volume
R ughly correlated ir some data sets;
not in others
L)nCh 1989, Wallace et al. 1996. Blake and Loiselle 2001,
Faaborg et al. this volume
Comlated in all habitats Wang and Finch 2002
ften correlated, bu not in all -.pecies
C rrelated in 73% of 64 species
Corresponded only roughly
Corresponded only roughly
Corresponded in 15 of 21 species
01 en corresponded
Of .en corresponded
( orresponded in -l of 4 '>pecies
orresponded in I of 2 '>pecic s
orrcsponded in I of 2 species
Silkey et al. 1999, Peach et al. 1his volume
Dunn et al. this rnlume a
Simons et al. tins volume
Simon-. et al. tlm 1·olume
Peach cl al. 1998. this l'Ol11111e
Berthold this 1·0/11111e
I lagan et al. 1992. Dunn and Husscll 1995, Dunn el al 1997. Francis and Husscll 1998. Rimmer cl al thi\ 1·oh1111e
du Feu and McMeeking this l'Olume
ur and Geupel l 991b
Nur and Geupel l 993b
B1eeding Population model ' Corresponded Bart et al. 1999
Survivorship Breeding
Breeding
Breeding
S x ratio
Cupture rate Breeding
Resighting
(I species studied)
Corresponded (I -;pecics studied)
Nur ct al. thi.\ rnl11111e
Band recoveries Corresponded roughly Peach and Baillie thi\· 1·olume (5 species studied)
Correlation with l vcral examples Peach et al. 1991. 1999 causal factor
hooting
Other trap t)pcs
o cnrre<,pondence (2 species)
Does not ah•ays correspond
Mawson 2000
Bauchau and Van oord\.\ijk 1995, Collister and Fisher 1995
moJcl ronta111ing rc,ulh Imm annual range "ide counh and annual 'un 1ul rates "a' u't'U to cst1111atc range "1ue prnduct1\ 11) 1n Kmland ·s Warbler (0 '11<hw rn kir1/a11dii) .
4 STUD IE I A VIAN BIOLOGY NO. 29
CES Scheme (Constant Effort Sites; Peach et al. this volume), the German MRI Program (Mettnau-Reit
Illmitz-Program; Kaiser and Berthold this volume), and the TOC program in France (Suivi Temporel du niveau d'abundance des populations d'Oiseaux terrestres Communs; Yansteenwegen et al. 1990). An evaluation of CE producti ity indices (Peach et al. 1996) showed that although there was variation in
capture rate and age proportions among locations. annual change in age proportions at individual tations were similar in direction and magnitude acros,
habitats and regions (Peach et al. 1996). Productivity indices ba ed on pooled data al. o were similar among a cluster of stations in California (Ralph et
al. this volume b), and pooled data from CES stations had acceptably low standard errors (Peach et al. thi. volume).
Migration data also may be useful for tracking regional productivity, as represented by the proportion
of young birds in fall mist-net samples. However, this hypothesis has been littl t sted (Hussell this volume). It will be difficult to validate productivity indices that are based on capture of fall migrants,
becau. e independent productivity data from the breeding grounds will rarely be available (because breeding locations ar unknown or unstudied). Nonetheless, some approaches to evaluation have been suggested for future research (Dunn ct al. this volume h).
MAPS, CES, and the other cooperative demographic monitoring programs mentioned above are designed to collect information not only on productivity, but also on apparent survival rates . Whereas
'-llfVival mt s could he e'>timated for any season in which birds are site faithful and relatively sedentary, these cooperative studies estimate annual survival betw en bre cling seasons. verage survival can abo be estimated for individual netting stations, although sample sizes are usually too low to document annual differences (Faaborg and Arendt 1995, Hilton and Miller 2003).
There are fewer validation studies of survivor-hip estimate. than of productivity indices, because
independent e timates of survivorship are harder to obtain. Nur et al. (lhi'i 1•olume) showed that survivor
ship of one species e. timated from mist-net recap
tures was similar to estimates based on re. ighting of marked individuals. Peach and Baillie (this volwne) found that acros. five . pecie , there was an overall (but non-significant) relationship between survivorship estimates based on CES and those based on band recoveries. Survival rates from CES were lower,
probably because birds that emigrate from a station cannot be di tinguished from birds that die, but the
authors presented cogent arguments ·upporting the usefulness of CES estimates as indices of survival. Th re hav also been se era! studie showing that change in annual survival rates was correlated with event. likely to have had a strong effect on mortality
(Peach et al. 1991, 1999).
POTENTIAL BIAS IN MIST-NET SAMPLES
A with bird counts obtained through visual and aural surveys, the numbers of birds captured in mist nets are indices of abundance, rather than total
counts. U ·e of standardized, constant effort protocol. will reduce variation in capture rates caused by un
even effort or net avoidance (Ralph et al. this volume a). However, e en completely standardized opera
tion capture only a proportion of all birds present. and that proportion will vary with specie., habitat, weather, and other factors unrelated to true popula
tion size. auer and Link (lhi.s volume) showed that capturing different proportions of the true population could lead to false conclusions in comparison of
samples, so it is important to investigate the potential for bias and to estimate its magnitude.
Capture rates at all seasons are affected by a multitude of factors, including distribution of nets with respect to territory size (Remsen and Good 1996, Ballard et al. this 1•olume, Nur et al. this \'Olume). mesh si1e of nets (Heimerdinger and Leberman 1966, Pardieck and Waide 1992, Jenni et al. 1996). season (Pagen et al. 2002), species (Jenni et al. 1996, Wang and Finch 2002), age class (Ballard et al. this 11ol11me, Burton and De ante this 1·0/z1mc>, Nur et al. this 11o l11111e), factors affecting movement rates (e.g ..
wh ther birds are incubating or molting), activity height (Remsen and Good 1996), and vegetation and habitat structure (Pagen et al. 2002, Ballard et al. this l'Ol11111c>, Kaiser and Berthold this 1•ol11mc>, Mallory et al. this 1·ol11me, Whitman this l'Olume).
apture rates of migrants are also affected by most of these factors. W ather ha:-. a particularly strong effect on migrant numbers, because it influence rate of daily influx and departure from a locati n, and weather effect may b especially marked at stations near the edges of migration routes ( imons et al. this volume). Jn addition, during migration there will be
daily variation in the proportion of birds migrating
past the study site that actually stop there (Dunn and Hussell 1995). Migrating birds may be less selective of habitat during migration than are breeding birds, however, so habitat biases may be lower during mi
gration than in other. easons. After a review of sources of bias in mist-net
captur s, Remsen and Good ( 1996) concluded that
MI T NETS AS A MONITORING TOOL-Dunn and Ralph 5
unadjusted capture rates should not be used in quantitative comparisons of relative abundance, either among species. or within species among habitats. On the other hand. there is much evidence that a strong signal can be obtained from standardized index counts (Table 1). Whereas descriptive, non-qualitative results alone can be useful for land managers (e.g .. Humple and Geupel 2002), information on relative abundance can add a great deal of value, particularly when conclusions are tempered by explicit discussion of the potential for bias and its possible magnitude. Moreover, long-term trend monitoring will not be compromised by the fact that numbers captured are only a proportion of true population -;ize. as long as there are no temporal trends in the capture proportions themselves. In most '>tudie. such stability is assumed rather than directly tested, but Dugger et al. (2000) found that capture proportions in a neotropical study area remained relatively stable over time within species and locations. However, relatively small change-; in a species' mean peak of activity can have a large effect on capture rates (Remsen and Good I 996). Long-term habitat change is the mo-;t likely source of systematic bias in longterm trend'> based on mi-;t netting (Ralph et al. this
\'O/ume a), and such change may be difficult to prevent even v, ith regular management of the vegetation (Kai-.er and Berthold this l'O!ume).
Mark-recapture methods can help to reJuce the potential for bia-. caused by variation in capture propmtions among mist-net samples ( aucr and Link thi.\ 1·0/11111<.>). Mark-recapture modeling estimates the proportion or all bird-; that i-. actually captured. v.hich can then he used to e..,timate total population si1.e (e.g., Kaiser and Bauer J 994, Kaiser and BcrtholJ 1111\· l'O!ume). Peach anJ Baillie (this
1•0/11111<!) and Kendal I et al. (this \'Ol11me) pro" ided background on the uses or mark recapture for this purpose. as well as for estimating adult survival, rcrruitment. and proportion or tr:rnsients in a sample. The technique ITI<\Y h<ne more limited value for migration -;tudies, because the high rate of turnover in the birds present at a study location precludes using recapture rates to estimate population siLe. It should be noted that capture-recaptur estimate-; of population size and capture probability are model-based, and the assumptions associated with any model must be considered when interpreting results.
Another mean'> of addressing bias that may exist in mist-net samples is to adjust numbers of birds captured according to independent data on abundance. Although no count methods are completely problem-free, a few technique. have been developed that produce relatively unbiased estimates
o1 density (Buckland et al. 2001, Bart and Earn st 2002, Thompson 2002). These methods can be used in combination with mist-netting studies to evaluate th~ presence and potential magnitude or bias in the m st-net amples. Once capture proportions have b en quantified, the density e. timation data can be u. d to adju t the mist-net sample. during analysi'>.
FUTURE RESEARCH
The strengths and limitations of mist netting for p pulation monitoring have received considerable att ntion in recent decades, but much remains to be learned. We suggest the following topics as priorities for research:
• The factors affecting the proportion of the true population captured need to be better quantified in a wider variety of species. In particular, more work is needed on effects of vegetation structure, habitat. and net avoidance.
• For programs that pool data from many stations. more work is needed on the most appropriate number and distribution of stations to en ure representative sampling at chosen geographic scales. the effects on re1.ults of frequency of operation. ,111d on cffecr... of '>t' tion turnover.
• Additional validation studies arc needed on abundance anJ demographic indices based on mi'lt netting (including fall age ratios in migrating bird'>), and on population trends of temperate migrants sampled in their wintering area .
•There is little information on age- or '>ex-speciric Jiffrrences in dispersal anJ habitat preference, or on detrree of annual variation in the..,c factor uch knowledge is important for interpreting spatial and temporal differences in productivity indices.
•Mark-recapture methods are improving rapidly. but better model'> are needed to address di persal of ju\cnilcs or previous breeders, and for pooling of data from 111ultip!P station'> (e'>pecially when there is turnover in the sample of stations). Use of mark-recapture for migration studies abo needs further im estigation.
CONCLUSIO S
Mist netting as an extremely valuable tool for many kinds of population monitoring, not only for detecting the presence of '>pecies and counting individuals, but as an efficient mean of capture to age individuals and mark them for future identification. It is almost unique among methods in providing demographic estimates in all seasons, for many species of birds. Although mi t netting i. e. pecially effective
6 STUDIES I A VIA BJOLOGY NO. 29
as a monitoring technique when used in markrecapture studie , it can also provide valuable indice of relative abundance. In addition, mist-net samples can be u ed to track long-term trends in abundance and productivity.
ACKNOWLEDGMENT
This paper benefited from discus.· ion among all au
thors and workshop participants (see list in preface.) We
especially appreciate the contributions of K. M. Burton, J. Faaborg, C. M. Handel , G. R. Geupel, ur. W. J. Peach ,
C. C. Rimmer, L. Thomas, C. Vansteenwegen, and A. A.
Whitman, and a review by C. M. Francis.